<![CDATA[The Aglaonema type and quality is difficult to identify due to leaf pattern variety. For this reason, a technique is developed to classify Aglaonema types from leaf images. The Aglaonema is identified using the Local Binary Pattern (LBP) technique. The LBP recognizes objects in the form of pixel neighbor patterns in binary code, which is sensitive to the radius (R) and the number of neighbors (P) pixels. In this article we will study the appropriate radius and number of neighbors so that the LBP code becomes an accurate abject texture attribute. Experimentally, R is varied from 1 to 5 while P is varied from 4 to 24 pixels. Two types of Aglaonema with two varieties taken from each type were used to test the accuracy of the LBP code. The accuracy of the classification results is carried out with the help of K-Nearest Neighbors (KNN). The results show that the greater the number of neighbors in determining the LBP code, the more accurate the classification results. Neighbors with a total of 18 have a stable accuracy reaching a total of 79%. Increasing the number of neighbors does not significantly affect accuracy. The neighbor radius affects the batik type of Aglaonema, the wider the neighbor area, the accuracy increases up to 84%, but for the Lipstick type, the best accuracy is obtained when R=3. By choosing the right R and P, the types of Aglaonema batik and Lipstick can be differentiated well.]]>
Copyrights © 2024
